Development/Plasticity/Repair

Myo-Cortical Crossed Feedback Reorganizes Primate MotorCortex Output

Timothy H. Lucas1 and Eberhard E. Fetz2

1Department of Neurosurgery, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania 19104 and 2Department of Physiologyand Biophysics, University of Washington, Seattle, Washington 98195

The motor system is capable of adapting to changed conditions such as amputations or lesions by reorganizing cortical representationsof peripheral musculature. To investigate the underlying mechanisms we induced targeted reorganization of motor output effects byestablishing an artificial recurrent connection between a forelimb muscle and an unrelated site in the primary motor cortex (M1) ofmacaques. A head-fixed computer transformed forelimb electromyographic activity into proportional subthreshold intracortical micro-stimulation (ICMS) during hours of unrestrained volitional behavior. This conditioning paradigm stimulated the cortical site for aparticular muscle in proportion to activation of another muscle and induced robust site- and input-specific reorganization of M1 outputeffects. Reorganization was observed within 25 min and could be maintained with intermittent conditioning for successive days. Controlstimulation that was independent of muscle activity, termed “pseudoconditioning,” failed to produce reorganization. Preconditioningoutput effects were gradually restored during volitional behaviors following the end of conditioning. The ease of changing the relation-ship between cortical sites and associated muscle responses suggests that under normal conditions these relations are maintainedthrough physiological feedback loops. These findings demonstrate that motor cortex outputs may be reorganized in a targeted andsustainable manner through artificial afferent feedback triggered from controllable and readily recorded muscle activity. Such corticalreorganization has implications for therapeutic treatment of neurological injuries.

IntroductionUnder normal behavioral conditions, primary motor cortex(M1) sites have a relatively stable bidirectional relationship withlimb muscles: stimulating a cortical efferent zone evokes consis-tent muscle responses (Asanuma and Rosen, 1972) and stimulat-ing muscle receptors activates neurons in the correspondingcortical zones (Rosen and Asanuma, 1972; Cheney and Fetz,1984). The stability of this reciprocal myo-cortical relationshipseems remarkable given abundant evidence that plastic changesof M1 movement representations can be induced by altered cir-cumstances (Sanes and Donoghue, 2000).

These consistent reciprocal relations within myo-corticalloops are normally maintained by the balance of synaptic inputsprovided through physiological pathways. However, altered cir-cumstances such as lesions can perturb the feedback conditionsand change the activation patterns of neuronal circuits. Consis-tent changes in the activation of these circuits can induce long-

term changes in synaptic strength, and such plasticity is thoughtto underlie cortical reorganization (Hebb, 1949; Markram et al.,1997; Cramer et al., 2011).

Examples of altered circumstances include the disruption ofnormal feedback pathways. Cortical reorganization can be pro-duced by selective disruption of pure motor nerves (Sanes et al.,1988), selective deafferentation (Kaas et al., 1983; Pons et al.,1991; Elbert et al., 1994), partial disruption of both motor andsensory pathways (Freund et al., 2011), central lesions (Nudo andMilliken, 1996), and complete loss of bidirectional communica-tion following amputation (Qi et al., 2000) or nerve division(Donoghue and Sanes, 1987).

Similar mechanisms are thought to underlie use-dependentplasticity during normal learning. Acquisition of new motor skillsinduces expansion of motor representations (Jenkins et al.,1990a; Pascual-Leone et al., 1995; Nudo et al., 1996; Hikosaka etal., 2002), just as sustained practice of sensory discriminationexpands sensory representations (Jenkins et al., 1990b;Recanzone et al., 1992a,b). Representational expansion precedesthe explicit phase of learning a new skill, demonstrating thatrapid functional plasticity of cortical outputs is associated withimplicit learning (Pascual-Leone et al., 1994). The expansion ofcortical representations, as measured by transcranial magneticstimulation (TMS), is accompanied by a decrease in the corticalactivation thresholds of muscles involved in the new motor task(Pascual-Leone et al., 1995). Moreover, mental rehearsal alonepromotes the modulation of neural circuits (Pascual-Leone et al.,1995), indicating that generation of movements is not a funda-mental requirement for motor cortical plasticity. These TMS ef-

Received Oct. 3, 2012; revised Jan. 18, 2013; accepted Feb. 7, 2013.Author contributions: T.H.L. and E.E.F. designed research; T.H.L. and E.E.F. performed research; T.H.L. and E.E.F.

analyzed data; T.H.L. and E.E.F. wrote the paper.Experiments were funded by the National Institutes of Health (NS12542, RR00166, E.E.F.; NS007144, T.H.L.), the

Washington Life Science Discovery Fund (E.E.F.), Washington Royalty Technology Transfer Fund (E.E.F. and T.H.L.),and the Congress of Neurological Surgeons Fellowship in Advanced Neurosystems Modulation (T.H.L.). We thank S.Perlmutter, A. Jackson, and C. Moritz for critical discussions relating to the experiments; L. Shupe for programmingassistance; G. Wu and Y. Nishimura for technical assistance; and G.A. Ojemann for support.

The authors declare no competing financial interests.Correspondence should be addressed to Timothy H. Lucas, 3400 Spruce Street, 3rd Floor, Silverstein, Hospital of

the University of Pennsylvania, Philadelphia, PA 19104. E-mail: Timothy.Lucas@uphs.upenn.edu.DOI:10.1523/JNEUROSCI.4683-12.2013

Copyright © 2013 the authors 0270-6474/13/335261-14$15.00/0

The Journal of Neuroscience, March 20, 2013 • 33(12):5261–5274 • 5261

fects are modulated in a manner consistent with mechanisms ofspike timing-dependent plasticity (STDP) (Wang et al., 1996).Changes in M1 during motor learning appear to involve long-term potentiation (LTP) of synapses (Rioult-Pedotti et al., 2000),mirroring plasticity mechanisms following injury.

Despite ample evidence of the capacity for plastic cortical re-organization, attempts to induce sustained motor output reorga-nization have been limited. We investigated whether and how therelationship between cortical efferent zones and forelimb mus-cles could be modulated by continuous activity-dependent feed-back from forelimb muscles to motor efferent zones of othermuscles. This was accomplished with a novel application of anautonomous head-fixed miniature computer called “neurochip”(Mavoori et al., 2005). Understanding the degree of flexibility ininteractions between cortex and muscles is fundamental to un-derstanding cortical control of movement. Targeted plasticitythat exploits this flexibility could provide novel strategies for clin-ical treatments of motor impairment.

Materials and MethodsSubjects and behavioral task. Four male Macaca nemestrina were trainedto perform a zero-torque wrist-holding task to provide consistent con-ditions for testing outputs evoked by intracortical microstimulation(ICMS). During training and testing sessions, animals sat quietly in aprimate chair with the right forelimb restrained in a neutral resting po-sition and the elbow bent at 90°. The hand was positioned around thecentral handle of a manipulandum that measured isometric torque aboutthe wrist in four axes (flexion– extension, pronation–supination, radial–ulnar deviation, grip–release). A molded clam-shell attachment securedthe hand within the device. A monitor in the booth displayed a trackingcursor for real-time feedback on the position and orientation of themanipulandum handle. Torques exerted upon the handle produced cor-responding movements of the cursor.

The task in this experiment required that the subject exert no forceon the manipulandum handle. A target box first appeared in the center ofthe monitor screen, corresponding to an isometrically neutral (zero-torque) position. The target box was only slightly larger than the trackingcursor, so any torques exerted upon the manipulandum handle resultedin a reset of the reward timer. The monkey was required to hold thetracking cursor within the target box for 3 s to trigger a reward of apple-sauce on a variable ratio schedule. Each trial was separated by a 3 s pausebefore the target box reappeared. Animals quickly learned to relax theforelimb muscles to keep the tracking cursor within the center target boxto attain the maximum frequency of rewards. This design ensured thatthe motor system was in a consistent state of baseline inactivation acrosstrials. This avoided any activation that would influence the corticallyevoked output effects. Animals were monitored continuously withclosed-circuit video. Trials were reset without reward if the animal failedto maintain the zero-torque position for the duration of the hold time.

Experimental paradigm. This experiment investigated the changes inmotor output effects produced by activity-dependent conditioning(ADC) between a forelimb muscle and an unrelated motor efferent zone.The experiment consisted of three phases: preconditioning “baseline,”conditioning, and postconditioning. Each phase involved testing ses-sions in the primate booth while the animal performed the holding task.Output effects evoked by trains of ICMS at motor cortex sites were doc-umented during testing sessions. Output effects were quantified withICMS train-triggered averages (TTA) of wrist torques and electromyog-raphy (EMG) of activated forelimb muscles (see below, Torque and EMGanalysis).

Surgical procedures and implants. For the cortical implant surgeries, theanimals received cephalexin (25 mg/kg), dexamethasone (1 mg/kg), andanalgesics (ketoprofen 5 mg/kg plus buprenorphine 0.15 mg/kg) andanesthetics (isoflurane 2%, dexmedetomidine 0.05 mg/kg plus ketamine5 mg/kg). A craniotomy was performed at the location of the forelimbarea of M1 (anterior 13 mm, lateral 18 mm). The location of forelimb M1was confirmed with forelimb movements evoked by cortical surface

stimulation. The dura was bonded to the pia with drops of cyanoacrylateto stabilize motion. A custom electrode array (Jackson and Fetz, 2007)was lowered into position above forelimb M1, such that the long axis ofthe microwires was oriented parallel to the bank of the precentral gyrus.Each microwire was manually advanced under direct visual guidanceuntil it penetrated the pia. The solid tungsten wires were sufficiently rigidto avoid deformation during penetration of the pia. The array was en-closed and anchored to the skull and encased in a titanium chamber withmethyl methacrylate. Animals were permitted to recover for 2 weeksbefore further procedures or experimentation. Animals received analge-sics on a regular schedule during this interval.

During subsequent procedures, M1 output effects from the electrodeswere documented under light ketamine sedation. Microwires were incre-mentally advanced down the bank of the precentral gyrus until evokedresponses were identified. Each array consisted of individually moveableTeflon-coated tungsten microwires (50 or 127 �m; geometric surfacearea 1.96 � 10 �5 cm 2 or 1.26 � 10 �4 cm 2) cut at right angles to produceimpedances ranging between 0.01 and 2 M� at 1 kHz (Jackson and Fetz,2007). The cortical surface area sampled by these arrays ranged between16.4 and 24.3 mm 2, centered on forelimb M1. Average interelectrodedistances ranged between 0.57 and 1.14 mm.

Short trains of ICMS delivered through each microwire activated con-tralateral forelimb muscles. Muscle activations were identified by visualinspection of fascicle contraction, manual palpation, joint movements,and EMG recordings via temporarily inserted needle electrodes. Musclesconsistently activated by ICMS were implanted with pairs of transcuta-neous EMG wires to record muscle activity chronically. EMG electrodesconsisted of insulated, braided stainless steel wires (31 American wiregauge). Two wires spaced 1 cm apart were passed into the muscle belliesof targeted muscles after bending the tip to create a hook and back-loading in 18 g hollow bore needles. Appropriate placement was con-firmed by evoking muscle contractions with low-threshold electricalstimulation through the EMG wires (0.5–2 mA, 30 ms train at 300 Hz)and by monitoring EMG responses on an oscilloscope during ICMStesting. The wires were secured on the skin with drops of cyanoacrylate.Medical grade adhesive tape provided additional support. The forelimbwas then wrapped in gauze padding for comfort. Finally, the forelimband torso were protected with custom canvas jackets to prevent dis-lodgment of EMG wires. Animals received scheduled analgesics dailywhile EMG implants were in place. All procedures were approved bythe Institutional Animal Care and Use Committee at the University ofWashington.

ADC. Three motor output sites and their corresponding activatedmuscles were selected for each experiment on the basis of stable anddiscrete output effects evoked by short ICMS trains with the arm restingin a neutral position. Motor cortex sites and their activated muscles weredesignated as follows: one cortical site (Nrec) activated the forelimb mus-cle (Mrec) chosen for recording during conditioning; a second corticalsite (Nstim), chosen for EMG-triggered stimulation during condition-ing, activated a different muscle (Mstim); a third cortical site (Ctrl) andits activated muscle (Mctrl), were chosen as internal controls (Fig. 1a).

During the preconditioning baseline phase, the output effects in mus-cles and torques evoked from the three sites were documented. Duringthe conditioning phase, activity of the Mrec muscle triggered stimulationof the Nstim cortical site via the continuous operation of a head-fixedcomputer, called neurochip (Mavoori et al., 2005; Jackson et al., 2006b).The neurochip was programmed as described below and activated at thebeginning of the conditioning phase. The animal was then returned tothe home cage. Throughout this phase the animal engaged in the usualrepertoire of unrestrained volitional behaviors in the home cage, in-cluding reaching, feeding, and grooming. The home cage was popu-lated with enrichment toys to encourage exploration andmanipulation with the forelimb under investigation. Small treatswere hidden within toys to encourage bimanual exploratory behav-iors. For overnight sessions, the neurochip remained active duringsleep. No abnormal forelimb movements or seizures were noted dur-ing conditioning in any of the experiments.

The neurochip contained the circuitry to record, digitize, and discrim-inate muscle activity and deliver recurrent intracortical stimulation in

5262 • J. Neurosci., March 20, 2013 • 33(12):5261–5274 Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1

real-time (Mavoori et al., 2005; Jackson et al., 2006b). The neurochiprecorded analog EMG signals that were bandpass filtered (500 –5 kHz)with a four-pole multiple feedback Butterworth filter, and then amplified�1500. After analog-to-digital conversion (at 11.7 ksps) the signalpassed to the first of two “Programmable Systems-on-a-Chip” (CypressMicrosystems). An 8-bit microprocessor core ran a signal discriminationalgorithm on the multiunit EMG activity. The algorithm first identifiedthreshold crossing and then discriminated signals that passed throughtwo adjustable time-voltage windows. Acceptance pulses were used as“triggers” that activated a control routine that ran a stimulator board at apredetermined stimulation ratio (defined as acceptance pulses: stimuluspulses). The neurochip was programmed to deliver subthreshold condi-tioning stimulation (43–56% of resting movement threshold) that didnot disrupt normal movements. Animals were returned to the home cageenvironment that allowed unrestricted movements.

The neurochip was tested in separate sessions for its ability to auton-omously record and discriminate multiunit muscle activity, as well asgenerate stimulation triggers for up to 30 h. The neurochip was pro-grammed to record 21 ms of raw data every 8.8 min to verify the qualityof EMG signal and discrimination. Stimulation rates were recorded forsuccessive 100 ms bins. Two additional channels of differential EMG(2 ksps) were also recorded. EMG data were bandpass filtered between20 –2000 Hz, with 250� amplification followed by full-wave rectifi-cation. Adjustable parameters to optimize signal detection and stim-ulation delivery included: DC offset, signal gain, detection threshold,discrimination windows (amplitude and delay), stimulation ratio,and stimulation intensity. Stimulation was disabled for a 4 ms refrac-tory period after each stimulus pulse to prevent artifact-triggeredstimulation during conditioning.

The conditioning phase concluded with a testing session in the pri-mate booth to document the output effects of Nrec, Nstim, and Ctrl. Theneurochip was deactivated immediately before this testing session. Sub-sequent postconditioning testing sessions documented the extinction ofany conditioned changes in output effects.

Experiments were designed to test the capacity for reorganization be-tween targeted pairs of M1 sites and the muscles that they activated. Three

of the five experiments tested the interactions be-tween forelimb “functional agonists” (Experi-ments 3–5; Table 1). Functional agonists weredefined as muscles that acted cooperatively dur-ing ethologically relevant behaviors, such as supi-nating the forelimb and extending the wrist, asmight occur when plucking a banana from a tree.Functional agonists exerted distinct actions at agiven joint when activated individually, andtherefore were not considered anatomical ago-nists. Two experiments tested interactions be-tween anatomical antagonists, namely musclesthat exerted opposing actions at a given joint, asin the case of extensors and flexors.

Torque and EMG analysis. Output effects ofthe cortical efferent zones were quantified withTTAs of wrist torques and forelimb EMG be-ginning 2 weeks after electrode implantation.ICMS thresholds for each site were determinedduring each experimental session. Thresholdswere defined as the lowest current intensitythat evoked wrist movement at least 50% of thetime while the animal performed the holdingtask. Selecting this current threshold meantthat up to half of the ICMS trains did not resultin overt movements. Stimuli consisted of a 39ms train of symmetric, negative–positive bi-phasic square wave pulses (0.2 ms phase �1) at333 Hz. Thresholds were 59.09 � 23.89 �A for50 �m wires (animals V, Y, and Z) and 96.67 �29.15 �A for 127 �m wires (animal A). Theestimated radius of tissue activated by directcurrent spread was 243 and 311 �m, respec-tively, based on the equation r 2 � i/k, where r is

the radius of activated tissue (in millimeters), i is current (in microam-peres), and k is a constant, assumed to be 1000 (Stoney et al., 1968).Polysynaptic effects recruited by temporal summation during stimulustrains are not limited to this radius (Jankowska et al., 1975).

Stimulus TTA of torque responses were compiled by averaging trialsover a peristimulus window (�500 ms). Full-scale torque responses werecalibrated at 0.5 N � m/5 V, and digitized at 5 kHz. TTAs at movementthreshold were transformed into 3D “torque space” (Fig. 2). TorqueTTAs were aligned on the center point origin in 3D plots, correspondingto the zero-torque position on all axes. This permitted serial comparisonsacross sites and experimental sessions. Curved lines represent meantorque trajectories evoked by ICMS at a site during an experimentalsession. Straight arrows represent mean torque vectors over the first 120ms of the torque trajectories.

TTAs were also generated at subthreshold and suprathreshold currentintensities (Fig. 3). Torque vectors and trajectories over a range of stim-ulus intensities at a single cortical site are illustrated. Torque vectors wereprojected onto the major planes of torque deflections (Fig. 3, arrowshadows) to permit angular measurements between torques evoked fromseparate cortical sites. Subsequent comparisons were made on the majorplanes.

Peristimulus TTAs of rectified EMG activity were compiled by aver-aging trials during the same interval as torque TTAs (�500 ms). Someraw EMG recordings showed ICMS artifacts, consisting of stereotypicwaveforms corresponding to the stimulus delivery times. Artifact correc-tion was performed using linear interpolation across 0.8 ms peristimuluswindow. Artifact correction was aided by the fact that ICMS stimulusfrequency, pulse width, time gate, and interstimulus interval were heldconstant throughout the testing periods.

EMG TTAs were used to calculate the mean normalized response,integrated area under the response curve, maximum response value, andresponse duration (Fig. 4). To identify responses, means and confidenceintervals (�3 SDs) of full-scale rectified EMG were calculated from the500 ms interval preceding stimulus onset. The “onsets” of evoked EMGresponses were determined with a sliding 6 ms detection window. Re-

Figure 1. Conditioning paradigm. a, ICMS testing before and after conditioning documented the muscle and torque responsesat movement threshold for different cortical sites with the forelimb in a resting, neutral position. Right, Example TTAs of rectifiedEMG and torque responses evoked from one cortical site. b, During ADC the neurochip discriminated Mrec raw EMG and deliveredproportional intracortical subthreshold stimuli at cortical site Nstim while the animal engaged in unrestrained behavior. Right,Mrec EMG activity exceeding threshold (horizontal dashed line)-triggered current pulses noted in the Nstim raster (blue).

Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1 J. Neurosci., March 20, 2013 • 33(12):5261–5274 • 5263

sponses were defined by values exceeding 3 SDsabove the baseline mean. The onset of such re-sponses was defined as the beginning of thesliding window. Similarly, EMG response “off-set” was defined as the time that the TTA fellwithin the confidence interval of baseline for 6ms (Fig. 3). Correspondence between TTA ofEMG and torque data is illustrated in Figure 5.Significant responses were compared beforeand during conditioning. One-way ANOVAwas performed on the mean normalized EMGresponses for each muscle during the responseenvelope, grouped by experiment phase. Datawere normalized by stimulus intensity to con-trol for changes in ICMS movement thresholdsbetween sites and across sessions.

Experimental controls. Conditioning ses-sions included control sites to determinewhether the conditioning effect was corticallysite specific (i.e., occurring only at Nrec) and input specific (i.e., occur-ring only during ADC). Within-session controls consisted of a neighbor-ing Ctrl site that was located in close proximity to Nrec and Nstim (Fig.6). ICMS at the Ctrl site was performed during each testing session todetect any nonspecific changes in M1 output effects attributable to stim-ulation alone, or to global changes in excitability or motivational states.The distances between the cortical entry sites were measured to deter-mine whether a spatial bias was present that would influence currentspread during conditioning or ICMS testing. There were no significantdifferences in the separation between pairs of cortical sites (Nstim andCtrl, 1.9 � 1.79 mm; Nstim and Nrec, 2.68 � 0.92 mm; and Nrec andCtrl, 3.11 � 1.44 mm; ANOVA, F(2,14) � 0.914, p � n.s.).

Additional controls were performed during “pseudoconditioning”sessions to determine whether the conditioning effects were input spe-cific. Pseudoconditioning consisted of stimulation delivered to Nstimsites that demonstrated the capacity for changes in motor output effectsduring ADC from Mrec. Pseudoconditioning sessions were interleavedin block design with ADC, either following or preceding conditioningsessions. Stimulation parameters for pseudoconditioning were matched

to ADC across the following parameters: location (Nstim), stimulationintensity, polarity, pulse width, mean frequency, total number of stimuli,duration, and behavioral state. Due to the limited programming capacityof the neurochip computer, pseudoconditioning was delivered at con-stant frequency, without the phasic modulation observed during EMG-triggered conditioning. Therefore pseudoconditioning had no consistenttemporal correlation with forelimb muscle activity. Similar methods ofdelivering matched stimulation, unassociated with forelimb movement,have been used to validate the requirement of activity dependence onmotor cortex reorganization (Porter et al., 2012).

Statistics. In addition to the statistics mentioned above, we computedsummary statistics. Statistical analyses between three or more indepen-dent groups with normal distributions were performed with two-tailedANOVA. Levene’s test for homogeneity of variance was performed toensure equal variance between sample groups. Two-tailed independentsamples T tests (t) were performed for comparisons between two groupswith normal distributions. To correct for cumulative Type I statisticerrors, Bonferroni’s post hoc test was performed for all ANOVA proce-dures. Groups that reached a familywise � value of 0.05 in post hoc testsare noted. ANOVA statistics are reported with the Fx,y � value, where x

Table 1. Conditioning stimulation parameters

Exp Figure Cx session

Conditioning parameters Direction of mean torque

Current (�A) Frequency (Hz) Duration (h) CE (°) Nrec (°) Nstim (°) Control (°)

1 7b 1 39 5.3 23.5 91.8 �13.3 �74.3 67.12 39 5.2 24.5 47.7 30.8 �71.6 51.83 39 7.4 25.4 43.6 34.9 �86.9 87.8

2 7e 1 39 7.3 5.9 78.2 232.4 288.9 53.42 39 2.7 9.1 52 206.2 278.1 89.63 39 4.3 12.1 74.6 228.8 253.1 54.54 39 10.3 9.8 64.7 218.9 347.9 74.75 39 9.8 9.8 96.7 250.9 304.2 66.7

3 8 1 44 9 0.4 46.7 231.2 204.3 176.52 44 9.1 22.8 58.5 219.4 215.4 165.53 44 6.7 24 96 181.9 192.5 176.84§ 44 9 0.4 17.7 260.9 205.7 170

4 9 1 39 4.2 3.1 59.1 �9 99.7 131.22 39 1.9 21 82.5 14.4 66.9 131.73 18 3.9 2.3 81.5 13.4 78.1 128.54 18 1.5 12.8 84.5 16.4 136.2 146.35 18 1.9 20 64.5 �3.6 113.9 150.26 18 2.2 21.2 59.4 �8.7 123.5 151.77 18 2.1 23.5 72 3.9 83.1 143.28 18 1.4 22.6 70 1.9 103.4 —*9 18 1 25.4 68.8 1.7 86.2 —

10 18 0.5 23.3 68.6 1.5 82.7 —5 — 1§ 39 5 0.3 1 178.1 124 162

2 39 8.4 0.3 175.1 4 138 1643§ 39 8 0.3 7.1 172 114 163

Conditioning parameters for five experiments are shown. Conditioning sessions are numbered sequentially. §, Pseudoconditioning sessions; *, loss of microwire due to mechanical failure. Cx, Conditioning; CE, conditioning effect.

Figure 2. Torque vector analysis. TTA plots show torques evoked by trains of ICMS stimuli, called torque TTAs, computed foreach manipulandum axis and aligned on onset of stimulus train (0 ms). Torque axes were combined to generate the torquetrajectory (curved line) and mean torque vector (green arrow) in 3D torque space for 120 ms following onset (*). Data representaverage of 41 ICMS responses at movement threshold evoked from a cortical site associated with forelimb pronation. Stimulusintensity 80 �A. Rad, radial deviation; Uln, ulnar deviation; Pro, pronation; Sup, supination; Flex, flexion; Ext, extension.

5264 • J. Neurosci., March 20, 2013 • 33(12):5261–5274 Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1

represents the degrees of freedom between groups and y represents thedegrees of freedom within groups. For each experiment, TTA data foreach cortical site were grouped by experiment phase (baseline, condi-tioning, and postconditioning), and factored on the direction of mean

torque per session for each cortical site. For summary statistics, dataacross all experiments were combined grouping by phase. Pearson’s cor-relation coefficients (r) were performed to test the strength of linearassociations between continuous variables. Nonparametric tests wereperformed with � 2 statistic (� 2) for categorical variables. Significance forall statistics was set at the 0.05 level. Error bars for all figures indicate 95%confidence intervals, unless otherwise stated. Additional statistical nota-tion is included, where appropriate, within the figure legends.

ResultsResults were obtained from four M. nemestrina. The data pre-sented here summarize 1025 ICMS recordings from 108 sessionscollected over a 24 month period. Of 108 sessions, there were 22conditioning sessions, comprising 5 conditioning experimentsand 2 pseudoconditioning experiments. ADC generated reor-ganization of motor output effects in all five conditioningexperiments.

Artificial connection between antagonist musclesRepresentative results for two conditioning experiments pairinganatomical antagonist muscles in separate animals are shown inFigure 7. In the first, conditioning-paired two-digit antagonists:activity of a flexor muscle (flexor digitorum superficialis � Mrec)triggered recurrent intracortical stimulation at a cortical site (Ns-tim) associated with an extensor muscle (extensor digitorumcommunis; EDC � Mstim) (Fig. 7a). Thus, conditioning createdthe following recurrent feedback loop: Nrec cortex3Mrecmuscle3Neurochip3Nstim cortex.

The mean baseline torque trajectories (curved lines) and av-erage torque vectors (straight line segments) for each cortical siteare shown in the plane that captured the dominant torque direc-tions (Fig. 7a, Baseline). Following 24 h of ADC, M1 outputeffects at these sites were documented (Fig. 7a, Conditioning).The Nrec torque vector (red) shifted 92° toward the Nstim vector(blue) following a single ADC session. Nstim and Ctrl vectorsremained stable, indicating that the conditioning effect observedat Nrec was site specific and not a generalized phenomenon dueto arousal, sensitization, or stimulation alone. The angular rota-tion of Nrec torque vectors, termed the “conditioning effect,” wasassociated with a substantial shift in the initial segment of thetorque trajectory (Fig. 7a, gray arrows), which was now alignedwith the Nstim torque. The stimulus intensity during ADC was39 �A (43% of resting movement threshold) and did not evokeany observed movements during conditioning. The stimulationrate triggered by muscle activity during this session fluctuated(Fig. 7c), with a mean rate of 5.2 � 4.5 Hz.

The direction of mean torque vectors across serial sessions isplotted in Figure 7b. ADC sessions are highlighted with shadedintervals. During each ADC session, Nrec demonstrated a robustconditioning effect (CE). The rotation of Nrec torque vector il-lustrated in Figure 7a is indicated on the plot of the direction ofmean torque with *. The mean CE across three 24 h conditioningsessions was 61.0 � 26.7° (Fig. 7b; Baseline vs ADC, t � 4.618, p �0.01). Mean output effects from the neighboring Ctrl site, sepa-rated from Nrec by 1 mm (Fig. 7 h), and the Nstim site deviatednegligibly from baseline after conditioning (t � �1.024, p � n.s.;t � 1.454, p � n.s.; resp.). The observed stimulation rate triggeredby muscle activity across all three ADC sessions fluctuatedsignificantly, with bursts of stimulation reflecting the profileof the muscle activity during free volitional movements (Fig.7c); the mean stimulation rate for all three ADC sessions was6.7 � 6.8 Hz.

Figure 3. ICMS torque responses as a function of stimulus intensity. Torque trajectories (coloredcurved lines) and mean torque vectors (colored straight lines) for different stimulus intensities deliv-ered to a cortical site associated with forelimb extension plotted in 3D torque space. Data represent120 ms following stimulus onset. To permit serial comparisons and angular measurements betweentorques, shadows of the torque vectors are projected (in gray) onto the major plane capturing thetorque directions. Subsequent comparisons are made on the major planes.

Figure 4. EMG Response to ICMS. TTAs of rectified EMG evoked during ICMS at two stimula-tion intensities in animal A. Response onsets were defined as times when the mean responseover a 6 ms sliding window exceeded 3 SDs above baseline mean. Baseline means were calcu-lated from the 500 ms immediately preceding stimulus onset (vertical line). Stimulus train isshown as horizontal dark gray bar. Higher stimulation intensity evokes larger TTA response(arrows). Red bars show baseline and response mean.

Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1 J. Neurosci., March 20, 2013 • 33(12):5261–5274 • 5265

Artificial connection betweenfunctional agonist musclesIn a second animal we created an artificialconnection between functional agonists,defined as pairs of muscles that contractedcooperatively during ethologically rele-vant forelimb movements, but did notproduce purely synergistic or antagonisticactions across a joint when contracting in-dividually. EDC (Mrec) activity triggeredstimulation of a cortical site (Nstim) thatactivated the forelimb supinator (Mstim).Conditioning over five sessions resultedin a robust and rapid shift in the evokedtorques of the Nrec cortical site (Fig. 7e–g). The mean CE was 73.3 � 16.6° in thedirection of Nstim (ANOVA, F(2,8) �33.094, p � 0.001, post hoc tests of condi-tioning vs baseline, p � 0.001). Meantorque directions evoked from the Ctrland Nstim sites did not change signifi-cantly across the conditioning block(ANOVA, F(2,8) � 0.245, p � n.s.;ANOVA, F(2,9) � 1.694, p � n.s.; resp.),although there was more variability in Ns-tim responses. The absence of a significantnet change in the mean torque at the Ctrlsite eliminated the possibility of modula-tion related to direct stimulation alone.The mean stimulation rate during ADCwas 6.9 � 8.9 Hz and the stimulus inten-sity (90 �A) was 43% of movementthreshold. The stimulation pattern washighly variable, reflecting variations inMrec EMG during unrestrained volitionalforelimb movements. The conditionedchanges in Nrec torque output returnedto baseline within 9.8 h after the end ofconditioning.

Requirement of activity-dependentstimulationTo test the hypothesis that activity-dependent stimulation was necessary and sufficient to inducereorganization in these experiments, we performed additionalcontrol experiments in which stimulation of Nstim unrelated tomuscle activity, termed pseudoconditioning, was alternated withADC. Pseudoconditioning was identical to ADC with respect tothe cortical stimulation site, current intensity, polarity, pulse-width, mean frequency, duration, absolute stimulus counts, andbehavioral condition—with one exception— unrelated stimula-tion was delivered at a constant frequency, with no correspon-dence to the activity of the forelimb muscles.

An example of these control experiments is illustrated in Fig-ure 8. Successful conditioning was first established by associatingsupinator muscle activity with stimulation of an M1 site thatactivated the digit extensor, EDC. A significant conditioning ef-fect of 46.7° was observed within 25 min of the onset of ADC (Fig.8a,*). By the second and third ADC sessions, the direction ofNrec and Nstim vectors did not differ statistically. The mean CEwas 67.1° (Baseline vs ADC, ANOVA, F(2,6) � 14.119, p � 0.005).The CE was then extinguished in the presence of normal sponta-neous activity without artificial feedback during unrestrained

movements. During pseudoconditioning PC, using stimulationrates identical as in *, no reorganization was observed (Fig. 8a,arrow; Baseline vs PC, t � 1.437, p � n.s.). Additional experi-ments interleaved ADC between blocks of pseudoconditioning,with similar results (Table 1).

Effective conditioning parametersWe tested the limits and sustainability of plasticity by switchingartificial feedback “on” and “off” over repeated conditioningblocks in a separate animal (Fig. 9). A robust conditioning effectof 59° was documented following a 3.2 h initial conditioningsession (Fig. 9b,*). This shift was accompanied by a significantchange in the direction of the initial segment of Nrec torquetrajectories (Fig. 9a, arrow) and a reversal of the direction of thetrajectory trace at its inflection point (Fig. 9a,†, Conditioning).To examine the effect of recruiting fewer cortical neurons duringconditioning, the stimulation intensity during the conditioningblock, ADC-1, was decreased from 39 to 18 �A (56 and 26%movement threshold, resp.) following the second session. Nochange in CE during the subsequent three sessions was noted,

Figure 5. ICMS responses from a cortical site associated with forelimb extension. Muscle responses and torque responses areshown for different stimulus intensities delivered at a single cortical site in animal A. Torque responses, shown at the right, increasewith stimulus intensity beginning at movement threshold (120 �A). Small muscle responses were observed at subthresholdstimulation intensities but failed to generate consistent torques. Horizontal gray bars represent stimulus trains and train onsetshown as vertical line. Calibration lines represent 50 mV (EMG) or 0.05 N � m (torque). ED 4/5, extensor digitorum digits 4 and 5;EDC, extensor digitorum communis; FDP, flexor digitorum profundus; Rad, radial deviation; Uln, ulnar deviation; Pro, pronation;Sup, supination; Flex, flexion; Ext, extension.

5266 • J. Neurosci., March 20, 2013 • 33(12):5261–5274 Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1

effectively ruling out extensive local current spread as an expla-nation of reorganization of output effects. The mean CE duringthe first conditioning block was �70° (difference of mean torquedirections during conditioning, �68.1 � 6.9°; direction of meantorque before conditioning, 6.3 � 11.7°; t � �11.1, p � 0.001).The CE was reduced with spontaneous activity in the absence ofartificial feedback within 20 h.

A second block, ADC-2, tested the sustainability of the CE byintroducing intermittent conditioning with a decreasing inter-mittent reinforcement schedule after the CE had been fully estab-lished. To achieve this, the ratio of intracortical stimuli-to-EMGtriggers during conditioning was decreased incrementally from1:4 to 1:16, while holding current intensity constant. The meanCE in block 2 remained highly significant (ANOVA, F(2,11) �7308.8, p � 0.001; Fig. 9), and was similar to the mean of the firstblock (CEblock 1, 74.4°; CEblock 2, 68.2°, Bonferroni post hoc test,p � n.s.). Interestingly, the CE following intermittent reinforce-ment persisted for one session 21 h after the end of ADC beforereturning to baseline levels.

Nstim movement thresholdsTo examine whether repeated subthreshold stimulation duringconditioning affected movement thresholds during subsequentICMS testing, Nstim thresholds were compared across experi-ment phases. In a previous experiment, movement thresholdswere followed for 60 d in a separate animal with a microwire arrayidentical to those used in the present study (our unpublished

observations). Thresholds were observed to fluctuate indepen-dently over time. To control for this variability, Nstim thresholdswere normalized by baseline thresholds for each experiment.There were no differences in the Nstim mean normalized thresh-olds among baseline, conditioning, postconditioning, andpseudoconditioning phases (ANOVA, F(3,45) � 1.141, p � n.s.).

Relationship between conditioning effect andstimulation parametersTo determine which, if any, of the conditioning parameters pre-dicted the magnitude of the CE, correlation analyses were run onthe variables of stimulation frequency (Hz), stimulation intensity(�A), conditioning duration (h), and total number of stimula-tion pulses delivered (n). The CE was not correlated with rate ofstimulation per session (r � 0.147, p � n.s.). There was no linearassociation between the mean stimulation rate and the magni-tude of the CE. The mean CE for ADC was 74.2 � 26.1°, with arange of 43.6 to 175.1°. This mean effect was relatively consistentover a range of average stimulation frequencies, from 0.52 to10.29 Hz. The mean CE was compared with total number ofstimulations (n), the product of the mean stimulation rate, andduration of conditioning. No significant correlations were ob-served between CE and n (r � �0.175, p � n.s.). Similarly, the CEwas independent of the total number of stimuli delivered acrossall ADC sessions (r � 0.337, p � 0.07), beyond the minimumvalue of 5940 stimuli delivered in one 25 min session.

The CE was not correlated with the duration of individualconditioning sessions (r � �0.292, p � n.s.), or the cumulativeduration of conditioning across all preceding sessions. The con-ditioning durations per session ranged broadly, from 25 min to�24 h. Conditioning for only 25 min resulted in significant CE,consistent with the rapid onset of reorganization noted in otherplasticity studies (see Discussion). These effects were as robustas the CE observed after 24 h of conditioning. Thus, in thiscase, conditioning beyond 25 min conferred no additional CE.There were no identifiable trends relating CE and condition-ing duration.

The CE was analyzed as a function of stimulation intensityduring conditioning sessions. There were no significant differ-ences in the magnitude of the CE across different stimulationintensities (ANOVA, F(4,17) � 0.97, p � n.s.), which ranged from18 to 59 �A. This suggests that the shift in torque vectors mea-sured during CE was not related to differences in the spread ofcurrent at different stimulation intensities. The mean CE for thehighest stimulation intensities (59 �A, 78.7 � 16.4°) was nearlyidentical to the lowest intensities (18 �A, 74.2 � 9.3°) necessaryto maintain the CE. The rate of extinction of the CE was notcorrelated with the conditioning parameters of conditioning du-ration, stimulation intensity, or mean stimulation frequency dur-ing conditioning.

In addition to quantifying torque responses, we docu-mented the muscle responses evoked with ICMS trains at eachcortical site before and immediately following conditioningsessions in a subset of muscles. Significant changes in TTAs ofnormalized EMG responses were found across experimentalphases and between efferent zones. The direction of EMGchanges was very variable. Muscle changes were categorizedinto the nominal variables of increased, decreased, or un-changed. Compiling muscle responses by cortical site acrossall experiments revealed that Mrec responses evoked fromICMS at Nrec increased following conditioning (� 2 � 6.4, p �0.05). Responses from Nstim and Ctrl were more variable,with equal proportions of muscle responses increasing, de-

Figure 6. Location of microwire tracks during conditioning experiments. a, Schematic oflateral cerebral hemisphere. b, Experiment 1, animal Z. c, Experiment 2, animal Y. d, Experiment3, animal Z. e, f, Experiments 4 and 5, animal A. Experiment numbers correspond to tableentries.

Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1 J. Neurosci., March 20, 2013 • 33(12):5261–5274 • 5267

creasing, or remaining unchanged (� 2 � 1.6, p � n.s.; � 2 �1.5, p � n.s., resp.). Figure 7d illustrates changes in muscleresponses: after delivering conditioning stimuli to Nstim trig-gered from Mrec (FDS), ICMS stimulation of Nrec evokedsignificantly larger responses of the Mstim muscle (EDC)

(ANOVA, F � 6.671, p � 0.002). Conditioning was also asso-ciated with significant decreases in Mstim responses evoked byICMS at Nstim (ANOVA, F � 79.738, p � 0.001) and nochange in responses evoked from the neighboring Ctrl(ANOVA, F � 2.079, p � n.s.).

Figure 7. Motor cortex output reorganization during conditioning. a, Average wrist torque vectors (colored arrows) and trajectory traces (curved lines) plotted in torque space in the plane thatcaptured dominant torque directions during ICMS at three cortical sites. Nrec vector (red arrow) shifts 92°after the first 24 h conditioning session (* in b). The initial 50 ms segment of the Nrectrajectory (indicated by gray arrows) reverses direction following Conditioning, aligning with the trajectory of Nstim. F, flexion; E, extension; G, grip; R, release. b, Direction of mean torques over 11consecutive days. The mean conditioning effect during 3 d of conditioning (gray regions) was 61.0 � 26.7°. Dashed red line shows mean Nrec at baseline. Error bars indicate SEM. c, Rates of stimulitriggered by Mrec muscle activity during free movement during first conditioning session. Mean rate for all three conditioning sessions was 6.9 � 6.8 Hz. Data concatenated in 100 ms bins. d, Muscleresponses evoked by ICMS. Red arrows indicate the appearance of responses in EDC (Mstim) evoked by ICMS at Nrec following a single conditioning session (* in b). e– g, A separate conditioningexperiment in another animal. Following 5.8 h of conditioning, Nrec torque vectors shifted 78.2° toward Nstim responses (* in f). P, pronation; S, supination; F, flexion; E, extension. f, Meanconditioning effect was 73.3 � 16.6° across five conditioning sessions. g, Stimulation rates (over 1 min bins) during all ADC sessions. h, The relative location of microwires for the two animals. C.S.,central sulcus.

5268 • J. Neurosci., March 20, 2013 • 33(12):5261–5274 Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1

Summary of resultsIn sum, we performed 22 conditioning sessions in four animalsover a 24 month period. ADC, triggered by forelimb muscle ac-tivity, produced significant reorganization of M1 output effectsin five of five experiments (100%) and 21 of 22 (95%) sessions.The conditioning effect was site specific: it was observed solely atNrec sites, with no significant effect observed at Nstim or neigh-boring Ctrl sites (Fig. 10). The direction of mean torques evokedfrom Nrec shifted an average of 74.2 � 26.1° from baseline duringADC (ANOVA, F(3,48) � 49.436, p � 0.001), compared with14.1 � 10.5° for Ctrl (ANOVA, F(3,32) � 2.071, p � n.s.) and17.2 � 15.7° for Nstim (ANOVA, F(3,50) � 2.06, p � n.s.). Theconditioning effect was input specific: it was observed only afterstimulation synchronized to relevant muscle activity, but not af-ter pseudoconditioning, which lacked a temporal relationshipwith muscle activity (Fig. 10a, red bars). Data are summarized inTable 1.

The Nrec torque vectors evoked by ICMS converged on Nstimbaseline vectors during ADC, as reflected by a reduction in theangles separating Nrec and Nstim torques (Fig. 10b). Points lyingbelow the dashed line of identity represent ICMS effects thatshifted toward Nstim. All Nrec responses met this criterion (Fig.10b, �). The mean angular separation between Nrec and Nstimwas reduced after ADC across experiments: Nrec torques shifted

toward Nstim after conditioning (t � 6.825, p � 0.001; Fig. 10c),while Ctrl effects showed no statistically significant change frombaseline (t � 1.481, p � n.s.). Normalizing these torque shiftsagainst the separation of torque directions at baseline reveals thatNrec torques shifted 49.8% of the baseline angular separationbetween Nstim and Nrec, whereas Ctrl torques shifted 29.8% ofthe baseline angular separation between Nstim and Ctrl.

DiscussionThis investigation sought to induce targeted reorganization of themotor cortex using crossed myo-cortical feedback. EMG signalsfrom contralateral forelimb muscles triggered stimulation of M1sites associated with different muscles. We found that ADC wasnecessary and sufficient to alter motor cortex output effects.Myo-cortical feedback induced rapid, robust, and input-specificreorganization. Reorganization was evident within 25 min ofADC, and was sustained with intermittent conditioning for up to5 d. Reorganization was site specific (occurring only for the Nrecsite) and input specific (occurring only with ADC). Pseudocon-ditioning, matched on most stimulation parameters to ADC,failed to induce reorganization. Stimulation triggered from EMGof the Mrec muscles did not produce significant conditioningeffects from the Ctrl sites.

A previous study using a similar intracortical recurrent con-ditioning paradigm found that long-term plasticity could be pro-duced by ADC triggered from action potentials (APs) of a singlecortical unit recorded at an Nrec site (Jackson et al., 2006a). Thepresent investigation documents that muscle activity can be aneffective surrogate of associated cortical activity for inducing tar-geted reorganization of motor cortex. This implies that the tim-ing of activity in Mrec is sufficiently correlated to neural activityat the Nrec site to mediate conditioning. In the study of Jackson etal. (2006a), each spike at Nrec triggered a stimulus at Nstim. Incontrast, myo-cortical conditioning established a probabilisticassociation between Nstim and Nrec cortical activity. In the pre-vious study reorganization of motor output effects increased overseveral days of conditioning and the changes in motor outputeffects persisted for up to 10 d following the end of intracorticalconditioning. Employing an EMG trigger changes the neuronalelements that could be involved in conditioning, because dispa-rate populations of corticospinal neurons are known to contrib-ute to muscle activity. While the stimulus intensity was heldconstant for each conditioning experiment of Jackson et al.(2006a), we were able to investigate the intensity and stimulationratio of conditioning stimuli delivered to Nstim as independentvariables. Differences in reorganization produced by intracorticaland myo-cortical conditioning may be related to differences inthe neural circuits involved, the precision of the synchrony be-tween neural elements, and the duration and intensity of thissynchrony.

This investigation also introduced pseudoconditioning as acontrol to determine whether stimulation alone was associatedwith changes in motor output effects. Pseudoconditioning wasidentical to ADC with respect to the cortical stimulation site,stimulation parameters, and behavioral task, with the critical ex-ception that it lacked the time-locked, phasic relationship to theactivity of the forelimb muscles. Ideally, pseudoconditioningwould include phasic activity to more closely mimic ADC; how-ever, limitations in the computational capability of the neurochipprecluded complex stimulus playback. The absence of significantreorganization during pseudoconditioning underscores the im-portance of ADC in inducing reorganization of output effects.These findings are consistent with recent evidence that

Figure 8. ADC is necessary and sufficient to induce reorganization. a, Direction of meantorques in separate conditioning experiment. Conditioning for 25 min produced a 46.7° shift inNrec torque responses (*). The mean conditioning effect across three conditioning sessions(ADC) in block 1 was 67.1°. Torque responses returned to near-baseline levels (dashed red line)following termination of artificial feedback. In the second block Nrec underwent pseudocondi-tioning (PC) using identical stimulation parameters as in *, including the behavioral condition,conditioning duration (25 min), current intensity (44 �A; 49% of movement threshold), andmean frequency (9 Hz). The total number of stimuli administered during ADC and PC was nearlyidentical (1.39 � 10 4 and 1.35 � 10 4, resp.). No reorganization was observed during PC(arrow). All sessions were 24 h in duration, except as noted above. b, The observed stimulationrate during the initial conditioning session of block 1 was 9.0 � 0.8 Hz. c, The programmedstimulation rate during PC was 9.0 Hz.

Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1 J. Neurosci., March 20, 2013 • 33(12):5261–5274 • 5269

movement-triggered vagal nerve stimula-tion induces reorganization of representa-tional maps of forelimb M1 in the rodent,while neither vagal nerve stimulation norforelimb movements alone were associ-ated with M1 plasticity (Porter et al.,2012). The fact that the Ctrl sites showedno significant changes in output effectsdespite the delivery of phasic EMG-triggered stimulation confirms that phasicstimulation alone, without a temporalcontingency between Ctrl and Nstim ac-tivity, was insufficient. Nrec sites demon-strated robust changes in torque vectorsduring conditioning relative to Ctrl andNstim sites, which demonstrated no sig-nificant changes from baseline fluctua-tions in torque output. We conclude thatboth close temporal association (Jacksonet al., 2006a) and behavioral relevance(Jenkins et al., 1990b; Nudo et al., 1996;Elbert and Rockstroh, 2004) are critical toinducing cortical plasticity in the sensori-motor system.

Accompanying the differences in meth-odology and design of the two studies, weobserved differences in output effects andreorganization. Generally speaking, ICMS-evoked torque responses were found to bemore variable than in the Jackson study(2006a). In many instances, there were smallfluctuations in torque direction evoked byICMS at a given microwire site between ses-sions. These fluctuations increased at allsites during conditioning. In addition tothese fluctuations, Ctrl sites with torque vec-tors nearly opposite to Nstim vectors dem-onstrated small shifts in torque vectorsduring conditioning. This may be related toactivation of partially overlapping musclefields of cortical projects during condition-ing triggered by the Nrec muscle.

Movement thresholds evoked by Ns-tim remained largely unchanged duringconditioning. In four of five experiments,no changes were observed. One experi-ment showed an increase in mean nor-malized movement thresholds similar toincreases noted by Jackson et al. (2006a).Jackson et al. (2006a) observed that insome cases Nstim output effects appeared to be suppressed fol-lowing conditioning, but returned over the next few days. Thesefindings were accompanied by variability in the magnitude ofNstim output effects, although the torque directions remainedstable. This variability may be attributed to long periods of re-peated stimulation that may temporarily affect electrode charac-teristics, reduce tissue excitability (McCreery et al., 1986), orincrease local inhibition. Local changes at Nstim, however, can-not account for the reorganization of output effects of Nrec ob-served during ADC.

The results can be explained by the potentiation of connec-tions between Nrec and Nstim within the cortex (Hess andDonoghue, 1994; Hess et al., 1996; Rioult-Pedotti et al., 1998).

We hypothesize that weak synaptic connections between corticalsites, or their downstream projections, are strengthened by thearrival of normal presynaptic volleys at Nstim coincident withinduced postsynaptic depolarizations produced by movement-related stimulation (Fig. 11). Subcortical sites where this condi-tioning paradigm could induce changes are illustrated by the“interneuron” pathway in Figure 11d. Additional sites of conver-gence, not illustrated, include the cerebellum. After conditioning,ICMS at Nrec recruits pathways normally activated from Nstim,resulting in a shift in evoked wrist torques toward those evokedfrom Nstim. Further, ADC strengthens these connections to thepoint that intermittent reinforcement alone is sufficient to main-tain them over multiple days. In this context, the intermittent

Figure 9. Sustaining plasticity by conditioning anatomical antagonists with intermittent reinforcement. a, Torque responses atbaseline and following 3.2 h of conditioning in animal Z. Nrec mean torque vectors (red arrows) shifted 59° above baseline. Thisshift was explained by significant change in the initial segment of the torque trajectories (gray arrows), and reversal of the directionof torque trajectory at its inflection point (†). Data represent first conditioning session shown in b (*). b, Direction of mean torqueduring an experiment with two conditioning blocks. The first block (ADC-1) tested effects of local current spread by decreasingcurrent intensity (indicated by intensity of gray shading in calibration bar). The second block (ADC-2) tested the sustainability ofplasticity by introducing intermittent reinforcement with a decreasing reinforcement schedule. The Nstim stimulation ratio (indi-cated by the height of the calibration bar) was incrementally decreased by a factor of 4 while the stimulation intensity was heldconstant (18 �A). Intermittent reinforcement maintained cortical reorganization throughout block 2. Nrec baseline torque shownby dashed red line. Ctrl microwire was lost to mechanical breakage between blocks. c, Stimulation rates during conditioning block1 (left) and 2 (right). Stimulation rates were highly phasic, reflecting the animal’s unrestrained limb movements. During block 2,the stimulation rate was reduced in two steps (§) by increasing the ratio of EMG triggers-to-intracortical stimuli during condition-ing. Rate data plotted for 1 min bins.

5270 • J. Neurosci., March 20, 2013 • 33(12):5261–5274 Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1

stimulation “reinforces” the newly established conditioning ef-fect, despite the decreased probability of conditioning stimula-tion relative to EMG triggers.

Accumulating experimental evidence indicates that motorcortex plasticity involves mechanisms of STDP. The studies ofJackson et al. (2006a) and Rebesco et al. (2010) demonstratedthat effective spike-stimulus delays occurred within the 50 ms ofSTDP effects. Given the nature of the EMG signal we did not testthe effects of varying delays. However, our results do demonstratethat the relation between cells at cortical site Nrec and muscleactivity in Mrec is sufficiently tight to induce plasticity withinprimate motor cortex or other effective sites.

Horizontal connections within M1 are capable of synapticmodification, including LTP (Hess and Donoghue, 1994, 1996,1999). Motor cortex LTP requires activation of NMDA receptors(Hess et al., 1996) and dendritic Ca 2� influx (Hess, 2002) as aprincipal mechanism of STDP. Coincident postsynaptic APs andunitary EPSPs result in increased amplitude of subsequent EPSPs(Markram et al., 1997). Blocking NMDA receptors prevents plas-tic changes in EPSPs. The mechanisms that underlie the plasticchanges observed here are likely to be the same as those thatunderlie the changes observed following altered physiological cir-cumstances (Pascual-Leone et al., 1995; Nudo and Milliken,1996; Nudo et al., 1996; Rioult-Pedotti et al., 2000; Sanes andDonoghue, 2000; Jackson et al., 2006a). In vitro (Markram et al.,1997) and in vivo (Jackson et al., 2006a; Rebesco et al., 2010;Thabit et al., 2010) investigations of cortical plasticity supportthis hypothesis and are consistent with STDP. Additional inves-

tigation into the cellular mechanism of synaptic strengtheningfollowing ADC is warranted. Convergent sites of synapticstrengthening at subcortical, cerebellar, or spinal levels are allconsistent with the present data (Fig. 11d, red boutons).

The fact that the conditioned changes are extinguished in thepresence of normal spontaneous activity without intermittentreinforcement is also consistent with the role of STDP mecha-nisms. Conditioning protocols that induce cellular changes aretypically reversed within minutes to hours by normal spikingactivity (Fregnac et al., 1988; Ahissar et al., 1992; Hess andDonoghue, 1999; Mu and Poo, 2006). Representational bordersbegin to return to their baseline topology within minutes follow-ing stimulation-induced expansion of cortical movement repre-sentations (Nudo et al., 1990). Functional changes in musclecontraction force and EMG following conditioning fade within30 min of cessation of associative stimulation (Taylor andMartin, 2009).

Non-STDP mechanisms could also be involved in M1 reorga-nization. The rapid reorganization observed in the present studyoutpaces axon sprouting and synaptogenesis (Kleim et al., 1996),making these mechanisms unlikely candidates to explain reorga-nization observed within minutes. Another alternative is fluctu-ations in the pattern of local inhibitory and excitatory inputs,which transiently influence M1 output effects on a shorter time-scale. For example, when intracortical inhibition is removed withbicuculline, a GABA antagonist, repeated stimulation unmasksextant, high-threshold excitatory connections resulting in theappearance of reorganization (Jacobs and Donoghue, 1991).Temporary release of inhibition may also facilitate synapticstrengthening between neighboring cell populations (Artola andSinger, 1987; Hess and Donoghue, 1994). Repetitive suprathresh-old stimulation (8� threshold) can induce a transient blockage ofneural activity adjacent to the stimulating electrode (Asanumaand Ward, 1971). Repetitive high-frequency stimulation at lowerintensities (2� threshold) expands movement representations inrodent motor cortex when delivered constantly for 1–3 h in ananesthetized preparation (Nudo et al., 1990). However, neitherrepresentational expansion nor local changes in inhibitory inputsexplain the results observed here, because (1) the reorganizationof ICMS output effects were observed from the Nrec site, whichwas not the site receiving stimulation during conditioning; (2)pseudoconditioning failed to produce reorganization despitematched stimulation parameters; and (3) evoked output effectsfrom Nstim and Ctrl remained largely unchanged after condi-tioning. Finally, movement-triggered vagal nerve stimulation hasrecently been shown to result in expansion in M1 representa-tional maps, with the proposed mechanism related to synergisticeffect of plasticity-enhancing molecules, including acetylcholine,norepinephrine, serotonin, and brain-derived neurotropic factor(Porter et al., 2012). Future investigations targeting the neuro-pharmacology of activity-dependent plasticity are indicated.

The ease of changing the relationship between cortical sitesand associated muscles suggests that these relations are normallymaintained through physiological feedback loops. Indeed, re-cordings from cortical and spinal neurons in these pathways re-veal very similar response patterns that would reinforce theirunderlying synaptic connections (Fig. 11e). Corticomotoneuro-nal cells (CM) have the most direct connections to spinal mo-toneurons (Cheney and Fetz, 1984). Motor units in forelimbmuscles exhibit corresponding response patterns during motortasks (for review, see Fetz et al., 2002a). Normal movements areassociated with afferent discharge from muscle receptors that isproportional to EMG, as recorded in dorsal root ganglia (Fig. 11e,

Figure 10. Reorganization is site specific and input specific. a, Change in mean torque di-rection for each cortical site for all stages of the experiments. ADC (black bars) produced signif-icant reorganization of torque output at Nrec, but not Nstim or Ctrl sites. Pseudoconditioning(red bars) failed to induce reorganization at any cortical site. Pooled data from all experimentsin four animals; number of observations shown in each bar. Baseline variability in torque outputshown in white. Data represent mean � SEM. ***p � 0.001, ns, not significant. b, Summary ofADC results illustrated by the angular separation of Nrec and Ctrl ICMS effects from the directionof Nstim mean torque (inset). Nrec ICMS effects (�) fall below the dashed line of identity inevery instance, indicating a consistent shift toward Nstim after conditioning. c, Baseline andconditioned mean ICMS effects for Nrec (filled bars) and Ctrl (open bars). Conditioning resultedin reduced separation of torque responses between Nrec and Nstim (***p � 0.001, indepen-dent samples t test, t � 6.825). There were no significant differences between baseline andconditioning Ctrl means.

Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1 J. Neurosci., March 20, 2013 • 33(12):5261–5274 • 5271

DRG). This afferent activity also activates CM cells (Cheney andFetz, 1984), creating a recurrent corticomuscular loop. The sim-ilar response patterns indicate that elements throughout the re-current loop are coactivated in a way that would maintain thestrength of their synaptic connections. Central inputs to the CMcell, including upstream neurons driving the CM cell (pre-CMcells), mediate the activation of CM cells during volitional move-ments. Under normal conditions the consistent coactivation ofthe inputs from pre-CM cells and from the afferent pathwaymaintains their synaptic strength. During our EMG-triggeredstimulation these afferents are consistently activated in syn-chrony with another muscle, artificially strengthening the con-nection from that muscle’s cortical representation. Onceestablished, intermittent activation is able to maintain inputspecificity. However, after termination of the artificial activationthe conditioning effects are reversed by normal reciprocalconnections.

The conditioning effects would be expected to appear in cor-tically evoked muscle activity as well as changes in torque out-puts. However, train-triggered EMG responses from sampledmuscles were more variable than wrist torque responses, whichrepresent the net sum of forces exerted by wrist and hand mus-cles. The variability in muscle responses may reflect physiologicalfactors that could not be controlled directly. First, these experi-ments paired muscles with different actions (e.g., FDS, EDC,supinator) in an attempt to test the range of conditionable con-nections. Muscles chosen to trigger recurrent conditioning mayfall within the muscle field of precentral cells that have excitatoryand/or inhibitory influences upon motoneuron pools with ago-nist or antagonist actions relative to the trigger muscle. This in-trinsic connectivity complicates predictions about the expectedEMG changes in response to conditioning. Torque responses

capture the actions of all wrist muscles, and therefore serve as amore comprehensive measure of motor output changes than asampling of forelimb muscles. Second, muscle responses are in-fluenced by the state of excitation of the �-motoneurons, whichare affected by various descending motor systems including thecorticospinal, rubrospinal, vestibulospinal, and reticulospinalpathways (Lemon, 2008), as well as spinal reflex pathways andmuscle afferents (Komori et al., 1992). The net activational stateof motoneuronal pools and corticospinal circuitry during a rest-ing task is difficult to control given the absence of volitional be-havior. While we carefully omitted trials when subjects activelymoved or failed to attend to the task, it was not possible to pre-vent the subjects from occasionally moving or shifting attentionbetween trials. Changes in wrist position between trials may affectthe excitability of motoneurons by changing the muscle thresh-olds. Changes in corticospinal excitability, which accompanywrist threshold position resetting, are independent of changes inEMG activity (Raptis et al., 2010). The degree to which changes incorticospinal excitability contributes to the variability of muscleresponses during ICMS at rest warrants future study.

A second factor that potentially influenced the variability ofEMG responses is the fact that the EMG electrodes were appliedtranscutaneously for each recording session. The implanted mus-cles were identified by intramuscular stimulation, which is ulti-mately mediated by the stimulated nerves. Although this wassimilar for each session, the motor units recorded could havebeen sufficiently different to account for the variation in evokedresponses.

Directing neocortical plasticity through recurrent activity-dependent stimulation has clinical implications for effectivetreatment of stroke and brain injury through plasticity or neuralprostheses. Targeted plasticity has been proposed as a therapeutic

Figure 11. Proposed mechanism of reorganization. a, ICMS testing defines baseline responses. b, Nstim ADC evokes postsynaptic depolarization synchronized to presynaptic inputs from Nrechorizontal connections normally activity during movement (dashed red arrow). c, Following conditioning, ICMS at Nrec activates strengthened pathways to produce responses that incorporate Nstimprojections. Similar mechanisms could occur at other sites of convergence. d, Putative convergent sites for synaptic strengthening. The projections of separate motor cortex populations areillustrated in a hypothetical conditioning experiment pairing anatomical agonists FCR and FDP. A stimulating microwire terminates in the proximity of Nstim. ADC strengthens synapses shown in red(“x”) that mediate connections from Nrec during volitional movements. A proportion of the population is activated directly by focal current injection (yellow region). Subcortical and spinal sites ofconvergence are illustrated by red boutons (y and z). e, Corticomuscle loop. A schematic of the minimal corticomuscle pathways (left) and response patterns of involved neurons (right). CM cells havethe most direct, monosynaptic connections on spinal motoneurons (MN), although polysynaptic linkages via other corticospinal cells exist. The afferent pathway begins with Ia afferents whosesomas are located in the DRG. The predominant response type of these cell populations recorded during a ramp-and-hold wrist movement is shown at right (Adapted from Fetz et al., 2002b).Additional inputs to the CM cell, including central cells driving the CM cell, are represented by the pre-CM cell.

5272 • J. Neurosci., March 20, 2013 • 33(12):5261–5274 Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1

method of reversing pathological neural activity (Engineer et al.,2011). Because the reorganization observed here required inter-mittent stimulation, clinical application of targeted reorganiza-tion based on these findings would probably require chronicstimulation to maintain efficacy. Similar approaches are com-monly used in neurological patients, as in the case of deep brainstimulation (Lyons, 2011), vagal nerve stimulation (Privitera etal., 2002), spinal cord stimulation (Compton et al., 2012), andmotor cortex stimulation (Compton et al., 2012). In each of thesetreatments, chronic stimulation is required to maintain efficacy.Chronic centrally triggered brain stimulation is under investiga-tion as a treatment for seizure disorders. The Neuropace RNSsystem, for instance, records electrocorticography in real-timeand delivers recurrent stimulation “to normalize” brain activitywhen seizure activity detected (Morrell and Investigators, 2009).A novel divergence from these approaches would be the applica-tion of activity-dependent recurrent stimulation triggered by pe-ripheral physiological signals. Employment of peripherallytriggered recurrent stimulation offers the potential advantage ofnot requiring stable chronic neural recordings. Additional ad-vantages include calibrated volitional control of effects and po-tential strengthening of weakened corticomuscular linkages.Peripherally triggered activity-dependent stimulation could em-ploy the same implantable battery powered platforms currentlyused for deep brain stimulation. Additional potential advantagesof using EMG activity over cortical activity as a source for induc-ing targeted reorganization in clinical brain– computer interfaceapplications are that EMG activity is (1) readily accessible, (2) hasa favorable signal-to-noise ratio, and (3) may be recorded indef-initely without the need of invasive procedures.

It is also possible that more prolonged conditioning wouldresult in more prolonged structural changes in synaptic connec-tions (Cramer et al., 2011). Another strategy to promote longerlasting effects would be to introduce additional mechanisms topromote effective long-lasting plasticity, like cortical DC polar-ization or concurrent delivery of neuromodulators. These possi-bilities would be issues for further investigation.

ReferencesAhissar E, Vaadia E, Ahissar M, Bergman H, Arieli A, Abeles M (1992) De-

pendence of cortical plasticity on correlated activity of single neurons andon behavioral context. Science 257:1412–1415. CrossRef Medline

Artola A, Singer W (1987) Long-term potentiation and NMDA receptors inrat visual cortex. Nature 330:649 – 652. CrossRef Medline

Asanuma H, Rosen I (1972) Topographical organization of cortical efferentzones projecting to distal forelimb muscles in the monkey. Exp Brain Res14:243–256. CrossRef Medline

Asanuma H, Ward JE (1971) Patterns of contraction of distal forelimb mus-cles produced by intracortical stimulation in cats. Brain Res 27:97–109.CrossRef Medline

Cheney PD, Fetz EE (1984) Corticomotoneuronal cells contribute to long-latency stretch reflexes in the rhesus monkey. J Physiol 349:249 –272.Medline

Compton AK, Shah B, Hayek SM (2012) Spinal cord stimulation: a review.Curr Pain Headache Rep 16:35– 42. CrossRef Medline

Cramer SC, Sur M, Dobkin BH, O’Brien C, Sanger TD, Trojanowski JQ,Rumsey JM, Hicks R, Cameron J, Chen D, Chen WG, Cohen LG, deC-harms C, Duffy CJ, Eden GF, Fetz EE, Filart R, Freund M, Grant SJ, HaberS, et al. (2011) Harnessing neuroplasticity for clinical applications.Brain 134:1591–1609. CrossRef Medline

Donoghue JP, Sanes JN (1987) Peripheral nerve injury in developing ratsreorganizes representation pattern in motor cortex. Proc Natl Acad SciU S A 84:1123–1126. CrossRef Medline

Elbert T, Rockstroh B (2004) Reorganization of human cerebral cortex: therange of changes following use and injury. Neuroscientist 10:129 –141.CrossRef Medline

Elbert T, Flor H, Birbaumer N, Knecht S, Hampson S, Larbig W, Taub E

(1994) Extensive reorganization of the somatosensory cortex in adulthumans after nervous system injury. Neuroreport 5:2593–2597. CrossRefMedline

Engineer ND, Riley JR, Seale JD, Vrana WA, Shetake JA, Sudanagunta SP,Borland MS, Kilgard MP (2011) Reversing pathological neural activityusing targeted plasticity. Nature 470:101–104. CrossRef Medline

Fetz EE, Perlmutter SI, Prut Y, Seki K (2002a) Functional properties of pri-mate spinal interneurones during voluntary hand movements. Adv ExpMed Biol 508:265–271. CrossRef Medline

Fetz EE, Perlmutter SI, Prut Y, Seki K, Votaw S (2002b) Roles of primatespinal interneurons in preparation and execution of voluntary handmovement. Brain Res Brain Res Rev 40:53– 65. CrossRef Medline

Fregnac Y, Shulz D, Thorpe S, Bienenstock E (1988) A cellular analogue ofvisual cortical plasticity. Nature 333:367–370. CrossRef Medline

Freund P, Rothwell J, Craggs M, Thompson AJ, Bestmann S (2011) Cortico-motor representation to the human forearm muscle changes followingcervical spinal cord injury. Eur J Neurosci 34:1839 –1846. CrossRefMedline

Hebb DO (1949) The organization of behavior: a neuropsychologicaltheory. New York: Wiley.

Hess G (2002) Calcium-induced long-term potentiation in horizontal con-nections of rat motor cortex. Brain Res 952:142–145. CrossRef Medline

Hess G, Donoghue JP (1994) Long-term potentiation of horizontal connec-tions provides a mechanism to reorganize cortical motor maps. J Neuro-physiol 71:2543–2547. Medline

Hess G, Donoghue JP (1996) Long-term potentiation and long-term de-pression of horizontal connections in rat motor cortex. Acta NeurobiolExp 56:397– 405. Medline

Hess G, Donoghue JP (1999) Facilitation of long-term potentiation in layerII/III horizontal connections of rat motor cortex following layer I stimu-lation: route of effect and cholinergic contributions. Exp Brain Res 127:279 –290. CrossRef Medline

Hess G, Aizenman CD, Donoghue JP (1996) Conditions for the inductionof long-term potentiation in layer II/III horizontal connections of the ratmotor cortex. J Neurophysiol 75:1765–1778. Medline

Hikosaka O, Rand MK, Nakamura K, Miyachi S, Kitaguchi K, Sakai K, Lu X,Shimo Y (2002) Long-term retention of motor skill in macaque mon-keys and humans. Exp Brain Res 147:494 –504. CrossRef Medline

Jackson A, Fetz EE (2007) Compact movable microwire array for long-termchronic unit recording in cerebral cortex of primates. J Neurophysiol98:3109 –3118. CrossRef Medline

Jackson A, Mavoori J, Fetz EE (2006a) Long-term motor cortex plasticityinduced by an electronic neural implant. Nature 444:56 – 60. CrossRefMedline

Jackson A, Moritz CT, Mavoori J, Lucas TH, Fetz EE (2006b) The Neuro-chip BCI: towards a neural prosthesis for upper limb function. IEEE TransNeural Syst Rehabil Eng 14:187–190. CrossRef Medline

Jacobs KM, Donoghue JP (1991) Reshaping the cortical motor map by un-masking latent intracortical connections. Science 251:944 –947. CrossRefMedline

Jankowska E, Padel Y, Tanaka R (1975) The mode of activation of pyramidaltract cells by intracortical stimuli. J Physiol 249:617– 636. Medline

Jenkins WM, Merzenich MM, Recanzone G (1990a) Neocortical represen-tational dynamics in adult primates: implications for neuropsychology.Neuropsychologia 28:573–584. CrossRef Medline

Jenkins WM, Merzenich MM, Ochs MT, Allard T, Guic-Robles E (1990b)Functional reorganization of somatosensory representations within area3b of adult owl monkey after behaviorally controlled tactile stimulation.J Neurophysiol 63:82–104. Medline

Kaas JH, Merzenich MM, Killackey HP (1983) The reorganization of so-matosensory cortex following peripheral nerve damage in adult and de-veloping mammals. Annu Rev Neurosci 6:325–356. CrossRef Medline

Kleim JA, Lussnig E, Schwarz ER, Comery TA, Greenough WT (1996) Syn-aptogenesis and fos expression in the motor cortex of the adult rat aftercomplex motor skill acquisition. J Neurosci 16:4529 – 4535. Medline

Komori T, Watson BV, Brown WF (1992) Influence of peripheral afferentson cortical and spinal motoneuron excitability. Muscle Nerve 15:48 –51.CrossRef Medline

Lemon RN (2008) Descending pathways in motor control. Annu Rev Neu-rosci 31:195–218. CrossRef Medline

Lyons MK (2011) Deep brain stimulation: current and future clinical appli-cations. Mayo Clin Proc 86:662– 672. CrossRef Medline

Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1 J. Neurosci., March 20, 2013 • 33(12):5261–5274 • 5273

Markram H, Lubke J, Frotscher M, Sakmann B (1997) Regulation of synap-tic efficacy by coincidence of postsynaptic APs and EPSPs. Science 275:213–215. CrossRef Medline

Mavoori J, Jackson A, Diorio C, Fetz E (2005) An autonomous implantablecomputer for neural recording and stimulation in unrestrained primates.J Neurosci Methods 148:71–77. CrossRef Medline

McCreery DB, Bullara LA, Agnew WF (1986) Neuronal activity evoked bychronically implanted intracortical microelectrodes. Exp Neurol 92:147–161. CrossRef Medline

Morrell MJ, Investigators RP (2009) Results of a multicenter double blindedrandomized controlled pivotal investigation of the RNS system for treat-ment of intractable partial epilepsy in adults. Presented at the 63 rd AnnualAmerican Epilepsy Society Meeting, Boston, MA, December 4 – 8.

Mu Y, Poo MM (2006) Spike time-dependent LTP/LTD mediates visualexperience-dependent plasticity in a developing retinotectal system. Neu-ron 50:115–125. CrossRef Medline

Nudo RJ, Milliken GW (1996) Reorganization of movement representa-tions in primary motor cortex following focal ischemic infarcts. J Neuro-physiol 75:2144 –2149. Medline

Nudo RJ, Jenkins WM, Merzenich MM (1990) Repetitive microstimulationalters the cortical representation of movements in adult rats. SomatosensMot Res 7:463– 483. CrossRef Medline

Nudo RJ, Milliken GW, Jenkins WM, Merzenich MM (1996) Use-dependent alterations of movement representations in primary motorcortex of adult squirrel monkeys. J Neurosci 16:785– 807. Medline

Pascual-Leone A, Grafman J, Hallett M (1994) Modulation of cortical mo-tor output maps during development of implicit and explicit knowledge.Science 263:1287–1289. CrossRef Medline

Pascual-Leone A, Nguyet D, Cohen LG, Brasil-Neto JP, Cammarota A, Hal-lett M (1995) Modulation of muscle responses evoked by transcranialmagnetic stimulation during the acquisition of new fine motor skills.J Neurophysiol 74:1037–1045. Medline

Pons TP, Garraghty PE, Ommaya AK, Kaas JH, Taub E, Mishkin M (1991)Massive cortical reorganization after sensory deafferentation in adult ma-caques. Science 252:1857–1860. CrossRef Medline

Porter BA, Khodaparast N, Fayyaz T, Cheung RJ, Ahmed SS, Vrana WA,Rennaker RL 2nd, Kilgard MP (2012) Repeatedly pairing vagus nervestimulation with a movement reorganizes primary motor cortex. CerebCortex 22:2365–2374. CrossRef Medline

Privitera MD, Welty TE, Ficker DM, Welge J (2002) Vagus nerve stimula-tion for partial seizures. Cochrane Database Syst Rev CD002896.

Qi HX, Stepniewska I, Kaas JH (2000) Reorganization of primary motor

cortex in adult macaque monkeys with long-standing amputations.J Neurophysiol 84:2133–2147. Medline

Raptis H, Burtet L, Forget R, Feldman AG (2010) Control of wrist positionand muscle relaxation by shifting spatial frames of reference for motoneu-ronal recruitment: possible involvement of corticospinal pathways.J Physiol 588:1551–1570. CrossRef Medline

Rebesco JM, Stevenson IH, Kording KP, Solla SA, Miller LE (2010) Rewir-ing neural interactions by micro-stimulation. Front Syst Neurosci 4.

Recanzone GH, Merzenich MM, Jenkins WM (1992a) Frequency discrimi-nation training engaging a restricted skin surface results in an emergenceof a cutaneous response zone in cortical area 3a. J Neurophysiol 67:1057–1070. Medline

Recanzone GH, Merzenich MM, Jenkins WM, Grajski KA, Dinse HR(1992b) Topographic reorganization of the hand representation in cor-tical area 3b of owl monkeys trained in a frequency-discrimination task.J Neurophysiol 67:1031–1056. Medline

Rioult-Pedotti MS, Friedman D, Hess G, Donoghue JP (1998) Strengthen-ing of horizontal cortical connections following skill learning. Nat Neu-rosci 1:230 –234. CrossRef Medline

Rioult-Pedotti MS, Friedman D, Donoghue JP (2000) Learning-inducedLTP in neocortex. Science 290:533–536. CrossRef Medline

Rosen I, Asanuma H (1972) Peripheral afferent inputs to the forelimb areaof the monkey motor cortex: input-output relations. Exp Brain Res 14:257–273. CrossRef Medline

Sanes JN, Donoghue JP (2000) Plasticity and primary motor cortex. AnnuRev Neurosci 23:393– 415. CrossRef Medline

Sanes JN, Suner S, Lando JF, Donoghue JP (1988) Rapid reorganization ofadult rat motor cortex somatic representation patterns after motor nerveinjury. Proc Natl Acad Sci U S A 85:2003–2007. CrossRef Medline

Stoney SD Jr, Thompson WD, Asanuma H (1968) Excitation of pyramidaltract cells by intracortical microstimulation: effective extent of stimulat-ing current. J Neurophysiol 31:659 – 669. Medline

Taylor JL, Martin PG (2009) Voluntary motor output is altered by spike-timing-dependent changes in the human corticospinal pathway. J Neuro-sci 29:11708 –11716. CrossRef Medline

Thabit MN, Ueki Y, Koganemaru S, Fawi G, Fukuyama H, Mima T (2010)Movement-related cortical stimulation can induce human motor plastic-ity. J Neurosci 30:11529 –11536. CrossRef Medline

Wang H, Wang X, Scheich H (1996) LTD and LTP induced by transcranialmagnetic stimulation in auditory cortex. Neuroreport 7:521–525.CrossRef Medline

5274 • J. Neurosci., March 20, 2013 • 33(12):5261–5274 Lucas and Fetz • Myo-Cortical Crossed Feedback Reorganizes M1